Proposed 2006 HIV Consensus Estimates

1
Proposed2006 HIV Consensus Estimates

• Estimation of the size of BRPs
• Estimation of HIV prevalence in BRPs
• Projection of HIV incidence in BRPs

• Willi McFarland, MD, PhD
• Director of HIV Surveillance, San Francisco
  Department of Public Health

2
Presentation

• Projected new HIV infections for 2006
• Major changes since 2001
• The numbers
• Population size of BRPs
• HIV prevalence in BRPs
• HIV incidence in BRPs
• The process to get the numbers
• Data and calculations behind the numbers
• Remaining uncertainties, next steps
• If time thoughts on why we have the current
  trends

3
2006 HIV Incidence Numbers

• 2006 Total 977
• 2001 Total 1,084
• Modest decrease of 10 in incidence number
• Incidence rates down in almost all groups
• Range 9 to 54 decrease

4
2006 vs. 2001 HIV Incidence
MSM Rate Down by 21
T-IDU Rate Down by 54
MSM-IDU Rate Down by 44
FSM/F Rate NA
MTF Rate Down by 39
MSF Rate NA
MSF-IDU Rate Down by 9
Children Rate Same
F-IDU Rate Down by 54
5
Impact

• 10 decrease in new HIV infections overall
• 10 - 50 decrease in HIV incidence rates
• If rates of 2001 continued, we would have seen
  1,301 new infections in 2006
• Would have 33 more new infections
• Real progress towards 50 reduction by 2010
• Reversed rising trend from 1995 to 2001

6
HIV incidence, San Francisco
Rise
Fall
Continued trend
Resurgence
Nadir
7
HIV incidence, San Francisco
Rise
Fall
Stable rate
Resurgence
Nadir
8
HIV incidence, San Francisco
Rise
Fall
Reversal of trend
Resurgence
Nadir
9
HIV incidence, San Francisco
Rise
Fall
Continued downward trend?
Resurgence
Nadir
10
Other major changes since 2001

• 25 increase in population size of MSM
• With MSM-IDU now 63,577 (58,343 without MSM-IDU)
• 31 increase in population size of MSM-IDU
• Some real increase, some under-estimation last
  time
• Growth in number of MSM offsets decrease in HIV
  incidence rates
• 34 decrease in population size of MTF
• 47 decrease in population size of MTF-IDU
• Over-estimation last time
• Probably really grew, just over-shot last time
• Ranking shifts
• MSM-IDU now 2 (was MTF, now 3)
• MSF-IDU now 4 (was FSM/F-IDU, now 5)

11
Proposed 2006 HIV Consensus Estimates

• MSM (non-IDU)
• 58,343 population size
• 14,205 living with HIV (prevalence 24.3)
• 772 new infections (incidence rate 1.75)
• MSM-IDU
• 5,234 population size
• 2,196 living with HIV (prevalence 42.0)
• 79 new infections (incidence rate 2.58)
• Transgender (non-IDU)
• 1,434 population size
• 327 living with HIV (prevalence 22.8)
• 42 new infections (incidence rate 3.78)

12
Proposed 2006 HIV Consensus Estimates

• MSF-IDU
• 7,076 population size
• 954 living with HIV (prevalence 13.5)
• 31 new infections (incidence rate 0.51)
• MTF-IDU
• 449 population size
• 194 living with HIV (prevalence 43.2)
• 16 new infections (incidence rate 6.08)
• FSM/F-IDU
• 4,030 population size
• 423 living with HIV (prevalence 10.5)
• 18 new infections (incidence rate 0.51)

13
Proposed 2006 HIV Consensus Estimates

• FSM/M
• 8,999 population size at risk (325,801 adult F
  non-IDU)
• 298 living with HIV (prevalence 3.3 or 0.091)
• 12 new infections (incidence rate 0.13)
• MSF
• 2,489 population size at risk (266,555 adult M
  non-IDU, non-MSM)
• 82 living with HIV (prevalence 3.3 or 0.031)
• 5 new infections (incidence rate 0.19)
• Children
• 13 births to HIV mothers (8,579 total births)
• 27 persons living with HIV infected as infants
• 2 new infections
• Blood
• 29 persons living with HIV infected via blood
  product exposure
• 0 expected new infections (risk lt 1 in 3 million)

14
Approach (Philosophy)

• All data are potentially biased
• Any data are better than no data
• More data are better than fewer data
• Examined gt51 sources of data
• Biases more important than precision
• gt75 researchers, service or care providers,
  public health officials, epidemiologists provided
  data, consultation, or served on a panel
• Expedient (draw the line at best we can do)
• On-going, but must come to a decision
• Consensus or Modified Delphi process

15
HIV Consensus Process

• HIV Consensus Meeting or Modified Delphi
  Process

Choose median estimate, plausibility bounds
June, 2005
Gather all known data
Calculate number projected in each

• HIV/AIDS/STI case reporting
• Behavioral surveillance
• Epidemiological studies
• Programmatic data
• Research

• Population prevalence
• Multiplier method
• Components method
• Trend analysis
• Expert opinion, etc

Panels of experts
Revise estimates
Present to public

• Researchers
• Health officials
• Service providers
• Steering committee

• Based on biases discussed
• Based on new data
• Based on expert opinion

• HIV Community Planning Groups
• Official report (Prevention Plan)
• Internet
• Press

April, 2006
16
Approach (Methods)

• There is no gold standard to compare to
• There is no specific study
• Select convergence of estimates by different
  methods
• Some biases over-estimate, some under-estimate
• Median (middle estimate) less influenced by
  outliers
• Choose upper and lower plausibility bounds
• Not confidence intervals
• Plausible lowest and highest that still make
  sense
• The pieces are inter-linked, not independent

17
Method Su Doku

• Examples
• Prevalence cant be lower than known living
  HIV/AIDS cases
• Incidence number must fit population size X rate
• Most projected numbers must show up in cases
  after several years
• Population size, HIV prevalence, HIV incidence
  must be reconciled
• If not, something is wrong

18
Methods of Estimation

• Population-based surveys
• Multiplier method
• Components (add up pieces)
• Ratio method (Lieb et al)
• Trends in indicators
• Roll forward (new infections minus AIDS deaths
  from 2001 consensus estimates)
• Last resort, consider mistakes in 2001
• Trends since 2001 consensus estimates
• Last resort, consider mistakes in 2001

19
Examples of Methods

• Population-based survey example American
  Communities Survey
• Adult male population (2005 proj.) x Percent
  self-reporting as gay or bisexual 338,977 x
  21.2 71,863 MSM (including MSM-IDU)
• Roll forward
• Start with 2001 estimate, add projected new
  infections, subtract deaths each year
• 198 48 48 48 48 23 23 -18 15 305
  Female IDU living with HIV
• Multiplier method National Behavioral
  Surveillance ATS clients
• Number of MSM ATS clients in 2004 (9 mo.) /
  MSM reporting ATS in 2004
• 588 / 0.9 65,333 MSM (including MSM-IDU)

20
Examples of Methods

• Components
• MSM living with AIDS new HIV-non-AIDS
  diagnoses 23 Unknown/not reported4867 6218
  3314 14,408 MSM living with HIV (not
  including MSM-IDU)
• Ratio
• Ratio of 1 MTF-IDU 3.45 MTF
• 449 MTF-IDU x 3.45 1,549 MTF
• Trends
• Indicator trends among MSM
• 3 Up, 3 Level, 6 Down Net down

21
Population size of MSM (non-IDU)
Upper plausible
2001 consensus 46,800
Median
Lower plausible
22
Population size of MSM-IDU
Upper plausible 7,689
Blue Estimates from IDU studies Yellow
Estimates from MSM studies
Median of medians 5,234
Lower plausible 3,923
23
HIV Prevalence, MSM
Upper plausible 15,216 (26.1)
Median 14,205 (24.3)
Lower plausible 12,252 (21.0)
24
HIV incidence indicators, MSM
UP
LEVEL
DOWN

• Repeat tester incidence
• CTRPN cases
• New HIV/AIDS cases

• STARHS AHP
• STARHS CC
• Incidence BSS
• Discordant UAI HIV
• Discordant UAI HIV-
• Syphilis

• Rectal GC
• AIDS with STD
• No. of partners

Interpretation Decrease in rate from 2.2 per
year 1.75 per year (772 in number)
25
Population size of MTF-IDU
Upper plausible
2001 Consensus 840
Median
Lower plausible
26
Population size of MTF
Upper plausible
2001 Consensus 2,160
Median 1,434
Lower plausible
27
HIV incidence indicators, TG-IDU
UP
LEVEL
DOWN

• CTRPN
• Reported cases

• STARHS ATS
• STARHS CC
• Repeat testers

Interpretation Overshot and down from 2001 of 40
new infections
Scenario 1 Reported cases 7 (2.75 per year)
Scenario 2 CTRPN diagnoses 24 (9.42 per year)
Scenario 3 Median of two scenarios 16 (6.08 per
year)
28
Pause Questions, comments?

• Do you want to see the calculations for any other
  BRP?
• Population size
• HIV prevalence
• HIV incidence
• What numbers do not have face validity to you?
• What have we not considered?
• What sources of data have we missed?

29
Next steps

• Within BRPs (difficult)
• Population size by race/ethnicity
• Prevalence by race/ethnicity
• Incidence by race/ethnicity
• Within BRPs (difficult)
• Population size by age (youth)
• Prevalence by age (youth)
• Incidence by age (youth)
• Mini-meth consensus
• Revisions?
• Report

30
If HIV incidence is down, why?

• Rapid identification and response to the rise
  detected from 1995 to 2001
• Maximum HAART coverage after 2001
• Prevention for HIV-positives
• HIV testing
• Community-originated prevention strategies
• Serosorting and others
• Other specifically targeted programs
• Methamphetamine use
• Syphilis, STDs
• Other?
• Continuing prevention that works (e.g., NEP)

31
San Francisco AIDS incidence, mortality, and
prevalence by year, 1980-2004
32
HIV incidence and syphilis among MSM
PS syphilis cases per 100,000
HIV Incidence ()
33
Rise in STDs but level HIV incidence?

• Syphilis too rare to drive HIV epidemic
• HAART suppresses HIV transmission but not STD
  transmission
• HIV serosorting

34
Sexual risk behavior among MSM
Unprotected anal sex
Potentially serodiscordant UAI, HIV
Potentially serodiscordant UAI, HIV-
35
Among 5 cities, San Francisco had the highest
level of HIV testing
Gay men testing for HIV, 2004
Note Not available for Miami
36
Among 5 cities, San Francisco had the lowest
level of unknown HIV infection
Unknown HIV infection among gay men, 2004
37
Among 5 cities, San Francisco had the lowest HIV
incidence among gay men
HIV incidence among gay men, 2004
38
CTRPN vs. HIV/AIDS reporting (MSM)
In the mid-term 72 of cases reported 95 if
diagnosed
2001 Consensus Estimate for HIV incidence
39
CTRPN vs. HIV/AIDS reporting (MTF)
In the mid-term 15 of cases tested and
reported?!
2001 Consensus Estimate for HIV incidence